[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#632667: empty biarch -dev packages



On Sun, Jul 24, 2011 at 06:05:02PM +0200, Steve Langasek wrote:
> Here is an alternate solution for this bug.  It's a bit eye-crossing, in
> that it means directories such as /usr/include/sys will now exist *only* if
> the biarch libc6-dev-$foo is installed, but it does the right thing (tested
> on amd64, and barring bugs should work on the other archs too).

While I agree this is not as clean as I would have hope, it makes the
job,and it's not too dirty provided we replace the biarch packages by
the corresponding multiarch before the release.

That reminds me we should start thinking about how to deal with
multiarch from the archive point of view. For i386/amd64, we can already
the switch, but for other architectures we are supposed to create the
corresponding architectures. We probably don't want to have full
architectures, and it's probably the time to start thinking more
precisely about that. The discussion should probably involve the
ftpmasters, porters from the concerned architectures, and maybe the
release team.

To come back to the patch, it might be better to use symlinks if
possible, otherwise peoples are going to complain we are wasting their
disk space...

> This is one option for getting multiarch libc6-dev on its feet.  The other
> would be to have libc6-dev-$foo install to the multiarch path, and have
> libc6-dev for the corresponding architecture Conflict/Replace with the
> biarch libc6-dev-$foo on an ongoing basis.

I agree that this solution should be avoided as long as possible, though
I start to think we might be forced to do it eventually :(

> I tend to favor the first option represented here because long-running
> Conflicts/Replaces for packages that are in the archive in parallel are
> generally bad; but ultimately, either should do the job.
> 

-- 
Aurelien Jarno	                        GPG: 1024D/F1BCDB73
aurelien@aurel32.net                 http://www.aurel32.net



Reply to: